An Image Recognition Technique for The Quantitative Analysis of Abnormally Oriented Grains in Advanced Piezoelectric Material

This work introduces a novel image recognition technique for the automatic detection of abnormally orientated grains (AOGs) in piezoelectric and ferroelectric thin films, with a specific focus on highly doped Scandium Aluminum Nitride (ScAlN). The algorithm accurately identifies and measures the grains on the substrate with pixel-level precision, facilitating precise calculations of the average area and coverage of the grains. The technique provides a powerful tool for researchers and manufacturers to evaluate and improve the quality of piezoelectric and ferroelectric materials via statistical inference.